Gallium nitride-based power semiconductor devices are well known to have properties that are desirable for power applications. Most of the proposed structures are lateral conductive devices which have power electrodes and control electrodes disposed along the top surface of the devices. Just below the electrodes, a hetero structure of aluminium gallium nitride (AlGaN) and gallium nitride (GaN), charges are generated at a hetero interface due to spontaneous polarization and piezoelectric polarization so that sheet carrier density of 1×1013 cm−2 or higher is obtained with no intentionally added impurities. As a result a high current density heterojunction field effect transistor (HFET) can be implemented by using a two-dimensional electron gas (2 DEG) generated at the heterointerface.
Nitride semiconductor-based power transistors have therefore been widely investigated and developed, and an on-resistance as low as one tenth or less of a Si-based metal oxide semiconductor field effect transistor (MOSFET) and one third or less of an insulated gate bipolar transistor (IGBT) has been implemented in the fields that require a breakdown voltage of 200 V or higher (e.g., see W. Saito et al., “IEEE Transactions on Electron Devices,” 2003, Vol. 50, No. 12, p. 2528). In a nitride semiconductor device, the size of an active region can be made smaller than in a Si-based semiconductor device. Therefore, reduction in size of the semiconductor device has also been expected for the nitride semiconductor device.
In a conventional nitride semiconductor device, the size of the active region can be reduced to about one third to about one tenth of the size of the active region of a Si-based semiconductor device. However, since an electrode pad for connecting wirings occupies a large area, the size of the current carrying interconnect tracks of lateral nitride semiconductor devices cannot be reduced sufficiently because of electromigration issues.
For example, a nitride semiconductor device shown in FIG. 11 has a drain electrode pad 125 connected to drain electrodes 118, a source electrode pad 126 connected to source electrodes 117, and a gate electrode pad 129 connected to gate electrodes 119. In this case, the area required for the nitride semiconductor device is about three times as large as the area of an active region 130. It is possible to reduce the size of an electrode pad, but such reduction in size of the electrode pad is limited in view of the yield.
It is also possible to form an electrode pad over the active region. In a nitride semiconductor device, however, a channel through which electrons drift extends in a direction parallel to a main surface of a substrate. Therefore, not only a gate electrode but a source electrode and a drain electrode are formed over the active region. In a power device, for example, a voltage of several hundreds of volts is applied between the drain electrode pad and the source electrode. It is therefore difficult to assure insulation between the drain electrode pad and the source electrode with a normal interlayer insulating film.
Moreover, in the case where an electrode pad is formed over the active region in the multi-finger nitride semiconductor device as shown in FIG. 11, the electrode pad and an electrode formed right under the electrode pad need to be connected to each other through a special plug or via. It is therefore difficult to connect the pad and also to assure the flatness of the electrode pad.